利用粒子群優化估算最佳化實驗設計之互動式平台

Web-based Interactive Platform for Finding Optimal Designs Using Particle Swarm Optimization.

吳亭慧(Ting-Hui Wu)

指導教授 ： 王偉仲

國立臺灣大學/理學院/應用數學科學研究所/碩士(2017年)

https://doi.org/10.6342/NTU201702210

摘要

如今隨著實驗規模成長，所需成本也隨之增加，因此在實驗之前就先評估成本以及估算最佳化的實驗設計為當今盛行的研究議題，而在估算最佳化的過程中，這裡採用粒子群優化(Particle Swarm Algorithm)來達成快速又不為複雜的效果，本文章透過建立網頁形式的溝通平台，連結原始MATLAB程式以及網頁伺服器，使得大眾團體可以直接透過網際網路操作而不需要相關的程式背景。

關鍵字

實驗設計 ； 粒子群優化

並列摘要

Since the experimental costs are increasing, optimal experimental design has been widely used nowadays for a more efficient process.
Using some algorithms to generate optimal designs for different specific models is more effective than finding analytical results for complicated models. Therefore, we use particle swarm optimization (PSO) to generate the best result for model selected by users. Here we built up a web-based interactive platform for practitioners, which contains different types of optimal designs for different experimental problems, and the users don't need any programming skills but can operate easily. We will demonstrate the usage of PSO in solving different types of optimal designs through the website.

並列關鍵字

optimal design ； Particle Swarm Optimization

參考文獻

[4] O. Paquet-Durand, V. Zettel, and B. Hitzmann. Optimal experimental design for parameter estimation of the peleg model. Chemometrics and Intelligent Laboratory Systems, 140:36–42, 2015.

[5] P.Salvo,R.Smajda,V.Dini,C.Saxby,G.Voirin,M.Romanelli,andF.DiFrancesco. A d-optimal design to model the performances of dressings and devices for negative pressure wound therapy. Journal of tissue viability, 25(2):83–90, 2016.

[6] K. Smith. On the standard deviations of adjusted and interpolated values of an ob- served polynomial function and its constants and the guidance they give towards a proper choice of the distribution of observations. Biometrika, 12(1-2):1–85, 1918.

[7] W. Wang, R.-B. Chen, C.-C. Huang, and W. K. Wong. Particle swarm optimization techniques for finding optimal mixture designs. 2012.

[8] W. K. Wong. Comparing robust properties of a, d, e and g-optimal designs. Compu- tational statistics & data analysis, 18(4):441–448, 1994.